<AWT Dev> RFR: 8239819 XToolkit: Misread of screen information memory
philip.race at oracle.com
Sun Mar 15 01:50:49 UTC 2020
So the net result is hard-coding
+ params.add(BORDER_PIXEL, Long.valueOf(0));
Why do we need to set it at all if its invisible ?
And does that mean its hidden / overwritten or something else.
Also the code says BORDER, but below you wrote BACKGROUND.
On 2/22/20, 1:05 AM, Sergey Bylokhov wrote:
> Please review the fix for jdk/client.
> Bug: https://bugs.openjdk.java.net/browse/JDK-8239819
> Fix: http://cr.openjdk.java.net/~serb/8239819/webrev.00
> The XAWT for some native structures uses special java wrappers. These
> wrappers have to be created on top of the native pointer and provide
> access to the native fields directly w/o JNI methods.
> We have incorrect usage of such wrapper in the XToolkit class:
> Toolkit creates AwtScreenData wrapper on top of the
> But unfortunately, this native method returns the wrong pointer
> to getDefaultConfig():
> getScreenData() should be used instead of getDefaultConfig(), but
> since native
> data for default configuration is bigger than screen data we most of the
> time read some "random" information about visual info.
> This code was added 17 years ago to set the background color of the
> XIconWindow which is used as an icon in the dock/taskbar/minimized icon.
> Looks like nobody noticed since then that the color used as a border is
> incorrect, because this color is actually is invisible.
> I have confirmed this using different color variants and even checked
> mwm(kind of cde on linux). So decided revert that fix back.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the awt-dev